A comparison of three black-box optimization approaches for model-based testing

Abstract

Model-based testing is a technique for generating test cases from a test model. Various notations and techniques have been used to express the test model and generate test cases from those models. Many use customized modelling languages and in-depth white-box static analysis for test generation. This allows for optimizing generated tests to specific paths in the model. Others use general-purpose programming languages and light-weight black-box dynamic analysis. While this light-weight approach allows for quick prototyping and easier integration with existing tools and user skills, optimizing the resulting test suite becomes more challenging since less information about the possible paths is available. In this paper, we present and compare three approaches to such black-box optimization.